Full-stack AI cloud collaboration built on Nvidia DSX platform; partnership to expand into physical AI and robotics
SK Telecom and Nvidia are accelerating plans to build a gigawatt-scale AI factory, targeting its first operation in 2027.
SK Telecom is joining Nvidia's cloud partner program to strengthen its position in the AI infrastructure market, with ambitions to grow beyond South Korea into a leading AI cloud provider across Asia.
SK Group Chairman Chey Tae-won and Nvidia CEO Jensen Huang outlined the plans Sunday at a press briefing held at the Serin Building in Jung-gu, Seoul. The announcement follows an earlier meeting in Taiwan, where the two executives reviewed an AI infrastructure roadmap and reaffirmed group-level cooperation.
The AI factory, slated to begin operation in 2027, will be a next-generation data center built on Nvidia's DSX platform — going well beyond the conventional data center model of general-purpose computing and data storage.
To that end, SK Telecom is joining the Nvidia Cloud Partner program, a global ecosystem for high-performance cloud AI services, with the goal of achieving the lowest token costs and the highest performance per watt.
The Nvidia partnership is expected to form the backbone of SK Telecom's AI cloud business. SK Telecom will receive support for AI training and inference starting with Nvidia's Blackwell GPUs, and will also adopt the latest Nvidia Vera Rubin platform, scheduled for supply in the second half of this year.
Nvidia, for its part, plans to draw on SK Hynix's memory chips as well as SK Telecom's expertise in building and operating AI factories.
SK Telecom aims to scale its AI cloud business rapidly, expanding the AI factory step by step into gigawatt-scale infrastructure. The company's broader goal is to become Asia's leading AI cloud provider, offering AI-specialized services spanning training, inference and data processing.
The two companies are also launching joint research into AI factory design and development, extending their collaboration — previously focused on semiconductors such as high-bandwidth memory chips — across the full spectrum of AI infrastructure, including factory construction and operation.
Cooperation will also deepen in physical AI and robotics. Nvidia Omniverse, a digital twin technology introduced at Nvidia GTC Taipei on June 1, is already being used in SK Hynix's semiconductor manufacturing processes.
"Through our partnership with Nvidia, we have built full-stack AI infrastructure competitiveness spanning chips to data center operations," Chey said. "By jointly addressing GPU, memory chip and energy challenges, we will emerge as Asia's leading AI cloud provider."
"Telecom networks, which are evolving into national AI infrastructure, are the foundation of AI cloud," Huang said. "Through the Nvidia DSX platform, SK Telecom will be able to build large-scale AI cloud and deliver agentic AI, enterprise AI and physical AI to companies around the world."
ko@heraldcorp.com
